Using AI for Automated Tasks (visualgui.com)

In a recent project, Nguyễn Đăng Liêm explored the potential of AI for automating typographic tasks while converting a 66-page non-selectable PDF of Vietnamese idiomatic expressions into semantic HTML. The endeavor leveraged an enterprise AI platform at George Mason University, featuring multiple AI agents from OpenAI, Anthropic, and Meta. Despite initial hurdles with GPT-5.4 and Claude Sonnet 4.6 running into limits, the project eventually found success in transforming the PDF into text and HTML, managing to incorporate smart quotes and clean markup. This experiment highlights the growing significance of AI in streamlining tedious tasks in digital design and text processing, particularly in dealing with complex languages that include diacritics. While the process still required manual intervention—most notably in retyping Vietnamese diacritics—the use of AI greatly accelerated the initial heavy lifting. This resonates with the broader AI/ML community by showcasing practical applications of automation in creative workflows, illustrating both the current capabilities and limitations of AI technologies in handling nuanced linguistic requirements.